Quant from 2024-08-20 does not have add_bos_token set to false

#1
by ThiloteE - opened
GPT4All-Community org

See discussion at https://huggingface.co/Qwen/Qwen2-7B-Instruct/discussions/15
The model works just fine, but if negative feedback emerges, then the next requantization should follow up and set add_bos_token in the tokenizer_config.json to false.

GPT4All-Community org

Maybe also remove "bos_token": null,, but that would require more testing.

Sign up or log in to comment